Highly stable and efficient solid-state solar cells based on methylammonium lead bromide (CH₃NH₃PbBr₃) perovskite quantum dots

Abstract

Easy processability and high stability are key features of methylammonium lead bromide (CH₃NH₃PbBr₃)-based perovskite solar cells. The main focus of the present work was to fabricate and evaluate the stability of CH₃NH₃PbBr₃ quantum dot (QD)-based perovskite solar cells. We used an ex situ solution process to synthesize CH₃NH₃PbBr₃ QDs and then successfully fabricated mesoscopic solid-state perovskite solar cells. We also studied the influence of different CH₃NH₃PbBr₃ QD sizes and different hole-transporting materials (HTMs), 2,2′,7,7′-tetrakis-(N,N-di-4-methoxyphenylamino)-9,9′-spirobifluorene (spiro-MeOTAD) and poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA), on the solar cell performance. The size of the CH₃NH₃PbBr₃ QDs was controlled by the solution processing parameters. Our controlled results show that spiro-MeOTAD- and PTAA-based devices exhibited, respectively, an open-circuit voltage (V_OC) of 0.991 and 1.091 V and a current density (J_SC) of 11.68 and 12.05 mA cm⁻², which resulted in an average power conversion efficiency (PCE) of 7.35 and 9.44% under a standard 100 mW cm⁻² illumination without masking. Our best-performing cell, which contains the FTO/Bl-TiO₂/mp-TiO₂+CH₃NH₃PbBr₃ (~2-nm QDs)/PTAA/Au configuration shows the following results: open-circuit voltage (V_OC)=1.110 V, current density (J_SC)=14.07 mA cm⁻², fill factor=0.73 and an 11.40% PCE. Furthermore, the CH₃NH₃PbBr₃-based devices are stable for more than four months.

Introduction

Methylammonium lead halide (MAPbX) (X=I, Br or Cl)-based perovskite solar cells (PSCs) open new approaches for the fabrication of efficient and stable solid-state dye-sensitized solar cells. The pioneering work on alkali-metal lead and tin halides was performed by Wells.¹ The chemical formula of the compound that he used was CsPbX₃ (X=Cl, Br or I). However, 94 years later, D. Weber et al. successfully replaced cesium (Cs) with methylammonium cations (CH₃NH₃⁺) and studied various compositions of the first three-dimensional organic–inorganic hybrid perovskites by tuning their crystal structures and phases.^{2, 3} Specifically, MAPbBr₃ perovskite shows the cubic phase I (Pm3m), tetragonal phase II (I4/mcm) and orthorhombic phase IV (Pnma, Z=4) systems. This is a p-type semiconducting material with a direct band gap of 1.93-2.3 eV that corresponds to an absorption onset of ⩽550 nm, which makes this material an excellent light harvester.^4,5 After successful synthesis of CH₃NH₃PbI₃ perovskite quantum dots (QDs), Im et al. demonstrated a 6.5% power conversion efficiency (PCE) via the ex situ method.⁶ Degradation of perovskite in a liquid electrolyte was solved by Gratzel et al. in 2013 using solid-state PSCs with a 13% PCE.⁷

The perovskite degradation problem has been nullified by implementing solid-state hole-transporting material (HTMs), for example, 2,2′,7,7′-tetrakis-(N,N-di-p-methoxyphenyl-amine)-9,9′-spirobifluorene (spiro-MeOTAD).⁷ Thus, for the last 2 years, methylammonium lead halide perovskite semiconductors have enabled of low cost solution-processed photovoltaic technology. On average, the PCE is typically boosted to 16.6% (certified), with the highest reported efficiency of ~19.3% (uncertified) for a CH₃NH₃PbI_3−xCl_x perovskite with a planar geometry and without an antireflective coating.⁸

Perovskite materials with different compositions have been successfully synthesized using in situ, ex situ, solvent engineering, compositional chemical management and vapor deposition techniques. Recently, Snaith et al. synthesized a mixed halide perovskite CH₃NH₃PbI_3−xCl_x using the dual-source vapor deposition technique to fabricatea planar heterojunction PSC.⁹ An open-circuit voltage (V_OC) of 1.07 V (the highest reported so far), a high current density (J_SC) of 21.5 mA cm⁻², and a 15.4% PCE were achieved. These results were achieved using the pinhole-free perovskite layer deposition technique, which enables thermal evaporation with a uniform HTM layer and a high diffusion length (L_D; 1069 nm).^{9, 10} In addition, the V_OC of the perovskite material can be controlled with electron-transporting layers (ETLs) such as TiO₂, Al₂O₃ and Yttrium-doped TiO₂ (Y-TiO₂).^{8, 11} Y. H. Hu discussed the state of the art of novel meso-superstructured solar cells that are based on an insulating ETL (Al₂O₃) and a lead-free perovskite (MASnI₃) for solar cell applications.¹² Recently, we have synthesized CH₃NH₃PbI₃ perovskite nanoparticles from γ-butyrolactone solvent and successfully developed low temperature-processed one-dimensional TiO₂ nanorod arrays with a 13.45% PCE.^{13, 14} Low temperature flexible PSCs have also been demonstrated by Yang et al. using phenyl-C₆₁-butyric acid methyl ester (PCMB) as the HTM material, resulting in a 9.2% PCE.¹⁵ However, a uniform deposition and thickness optimization of the phenyl-C₆₁-butyric acid methyl ester is difficult to achieve.¹⁶

Recently, Seok et al. demonstrated a 16.2% PCE using solvent engineering and toluene drip casting treatment for perovskite materials.¹⁷ This treatment helps to form an intermediate MAI–PbI₂–dimethyl sulfoxide (DMSO) phase, which retards the rapid reaction between PbI₂ and MAI during the evaporation step and results in a highly pure crystalline CH₃NH₃PbI₃ perovskite layer. However, these devices suffer from a low fill factor (FF) and hysteresis issues. The optimization of each layer is a major challenge for MAPbI₃-based PSCs, but the hysteresis issue can be solved by varying the thickness of the mesoporous TiO₂ (mp-TiO₂) layer. Seok et al. studied different compositions of methylammonium lead iodide (MAPbI₃), formamidinium lead iodide (FAPbI₃) and methylammonium lead bromide (MAPbBr₃) as effective light harvesters for PSCs. From this study, they concluded that the incorporation of MAPbBr₃ into FAPbI₃ stabilized the perovskite phase of FAPbI₃ and improved the PCE of the solar cell. The authors¹⁸ demonstrated a >18% PCE for mixed (FAPbI₃)_1−x (MAPbBr₃)_x, where x=0, 0.05 or 0.15. Therefore, a study of the bare MAPbBr₃ perovskite will open new opportunities in the architecture design of bilayer PSCs. Moreover, a high FF and high stability are the most important key features of MAPbBr₃-based PSCs. It is well known that MAPbX₃ nanoparticles/QDs with sizes <10 nm enable the preparation of new device architectures that could further enhance solar cell performance and elucidate the perovskite operating mechanisms.¹⁹ Therefore, MAPbBr₃ QDs-based PSCs open new approaches towards the development of stable and efficient solar cells. The optical properties of MAPbBr₃ and MAPbI₃ were investigated by Tanaka et al.²⁰. It was determined that for MAPbBr₃, the exciton binding energy and the exciton Bohr radius were 76 meV and 20 Å, respectively; for MAPbI₃, the exciton binding energy and exciton Bohr radius were 50 meV and 22 Å, respectively.²¹ The absorption coefficients of MAPbBr₃ and MAPbI₃ are 10⁵ cm⁻¹ and >4.3 × 10⁵ cm⁻¹ (at 300 nm), respectively.

Cai et al. demonstrated a 3.04% PCE using poly[N-9-hepta-decanyl-2,7-carbazole-alt-3,6-bis(thiophen-5-yl)-2,5-dioctyl-2,5-dihydropyrrolo[3,4–]pyrrole-1,4-dione] (PCBTDPP) as the HTM.²² However, the V_OC can be tuned by the band gap and the HTM of the materials; the authors demonstrated a V_OC of 1.15 and 1.3 V for PCBTDPP and N,N′-dialkylperylenediimide as the HTMs, respectively. Recently, E. Edri et al. demonstrated a 0.72% PCE with a V_OC of 1.06 V using phenyl-C₆₁-butyric acid methyl ester as the HTMs²³ for MAPbBr₃ PSCs. Moreover, Seok et al. demonstrated band tuning by incorporating MAPbBr₃ (that is, Br doping) into MAPbI₃ to achieve a PCE of 12.3% and a V_OC of 1.13 V using polytriarylamine (PTAA) as the hole conductor.²⁴ The incorporation of MAPbBr₃ improved the V_OC and the current density from 0.87 to 1.13 V and from 5 to 18 mA cm⁻², respectively.

Im et al. fabricated a planar MAPbBr₃ hybrid PSC using P3HT, PTAA and poly-indenofluoren-8-triarylamine (PIF8-TAA) as the HTMs. The MAPbBr₃ perovskites were synthesized in a dimethylformamide solvent, and the influence of H₂O content was studied. The authors showed a 7.3%, 9.3% and 10.4% PCE with a V_OC of 1.09, 1.35 and 1.51 V for the P3HT, PTAA and PIF8-TAA HTMs, respectively.²⁰ To the best of our knowledge, there are no reports describing the synthesis of the MAPbBr₃ QDs with <3 nm sizes in either the dimethylformamide or DMSO solvents that are used for the solar cell application.

In the current study, we report the preparation of highly stable and high conversion efficiency mp-TiO₂-based solid-state PSCs using MAPbBr₃ QDs. We synthesized highly crystalline MAPbBr₃ QDs with different sizes by the ex situ process in a DMSO solution followed by crystallization on mp-TiO₂. Two different types of HTMs, spiro-MeOTAD and PTAA, were used to study the performance of the solar cell devices. Because of the formation of very small (~2–3 nm) MAPbBr₃ perovskite QDs, the current density was drastically increased to 14.07 mA cm⁻². This resulted in an 11.14% conversion efficiency at 1 sun illumination. The resolution of the hysteresis issue is also discussed in detail.

Materials and Methods

Device fabrication

Laser-patterned FTO-coated glass substrates (TEC-8, Philkington Group Limited, St Helens, UK, supplied by HANALINTECH, Yongin, South Korea) were ultrasonically cleaned in an alkaline aqueous solution, rinsed with deionized water, ethanol and acetone, and subjected to an O₃ treatment for 30 min. A 20–40-nm-thick compact TiO₂ layer was deposited on the substrates by first spin coating the TiO₂ precursor and then annealing the spin-coated precursor in air at 450 °C for 30 min. The precursor composition was as follows: 0.15 and 0.3 M commercial titanium diisopropoxidebis(acetylacetonate) solution (75% in 2-propanol, Sigma-Aldrich) that was diluted in ethanol (1:39, volume ratio). After cooling the substrates to room temperature, they were treated with the 0.04 M TiCl₄ aqueous solution for 30 min at 70 °C, then rinsed with deionized water and dried at 500 °C for 20 min.

Deposition of the mesoporous TiO₂ layer

The mp-TiO₂ layer was composed of the 20-nm-sized particles and deposited by spin coating a commercial TiO₂ paste (Dyesol DSL-18NR-T, Dyesol, Queanbeyan, NSW, Australia) diluted in ethanol at 4000 r.p.m. for 15 s. However, the thickness of the mp-TiO₂ layer was optimized using spin-coating speeds from 2500 to 5000 r.p.m. After drying at 125 °C, the TiO₂ films were gradually heated to 500 °C, baked at this temperature for 15 min and then cooled to room temperature. The mesoporous deposited film was again treated with TiCl₄.

Preparation of methylammonium lead tribromide (CH₃NH₃PbBr₃)

Methylammonium lead bromide (MABr) was synthesized by the dropwise addition of hydrobromic acid (HBr) (48 wt% in water, Sigma-Aldrich, St Louis, MO, USA) to a methylamine solution (aqueous, 40 wt%, TCI Chemicals, Tokyo Chemical Industry Co. Ltd., Tokyo, Japan) in an ice bath. The ice-cold solution was stirred for 2 h followed by the solvent evaporation using a rotary evaporator (95 mbar vacuum, 400 r.p.m. rotation) at 60 °C. The resulting white product was dissolved in ethanol and recrystallized using diethyl ether. The fresh fine white crystals were washed three times using diethyl ether and then dried in vacuum for 24 h. The resulting white solid product was used for the CH₃NH₃PbBr₃ synthesis.

The CH₃NH₃PbBr₃ (MAPbBr₃) precursor solution was prepared by dissolving equimolar amounts of MABr and lead bromide (PbBr₂) (99.999%, Sigma-Aldrich) in DMSO (Alfa Aesar, Ward Hill, MA, USA) at room temperature overnight. The prepared transparent solution was filtered by a syringe filter (Whatman GD/X PVDF pore size 0.45 μm). The clear transparent solution was dripped onto the top of the mp-TiO₂ film. The film was allowed to soak for 1 min, and then was spun at 1000 r.p.m. for 20 s and 4000 r.p.m. for 30 s. The spin-coated substrate was dried on a hot plate and used for HTM deposition.

The HTM was prepared using a previously reported procedure with a few modifications. We used two different types of HTM for this study. In total, 180 mg of 2,2′,7,7′-tetrakis-(N,N-di-p-methoxyphenyl-amine)-9,9′-spirobifluorene (spiro-MeOTAD, Merck, KGaA Darmstadt, Germany) was dissolved in 1 ml of chlorobenzene (99.8%, Aldrich) with the addition of 37.5 μl bis(trifiuoromethane)sulfonimide lithium salt (LiTFSI, 99.95%, Aldrich) (170 mg ml⁻¹) in acetonitrile and 17.5 μl 4-tert-butylpyridine (96%, Aldrich). For the poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA)-based HTM, 15 mg of PTAA (EM Index, M_w=17 500 gmol⁻¹) in toluene (1.5 ml) was mixed with 15 μl solution of lithium bistrifluoromethanesulfonimidate (170 mg) in acetonitrile (1 ml) and 7.5 μl 4-tert-butylpyridine. The mixture was spin coated on the FTO/Bl-TiO₂/mp-TiO₂+MAPbBr₃ substrates at 3000 r.p.m. for 30 s. Then, the substrates were transferred to a vacuum chamber and evacuated to 2 × 10⁻⁶ mbar pressure. For the counter electrode, an 80 nm Au layer was deposited on top of the HTM layer using thermal evaporation (growth rate ~0.5 Å s⁻¹). The active area of the electrode was fixed at 0.09 cm². The active area was calculated using the cross-sectional area of the gold and of the laser pattern.

Results and Discussion

The crystalline structure of the synthesized MAPbBr₃ thin film was characterized by X-ray diffraction (XRD) analysis. Figure 1a shows the XRD pattern of the synthesized MAPbBr₃ thin film deposited on a glass substrate. The obtained XRD pattern is consistent with previously reported patterns in the literature.^{15, 16} The XRD pattern of the MAPbBr₃ sample showed that it had a highly crystalline cubic phase. The lattice parameter of the corresponding cubic phase of the MAPbBr₃ perovskite was found to be a=5.9334 (5) Å (a=5.94173 Å, space group=Pm3m).²⁵ The Bohr radius of CH₃NH₃PbBr₃ is 20 Å.^{26, 27} Figure 1b shows a typical photoluminescence (PL) spectrum of MAPbBr₃. The MAPbBr₃ film exhibits an emission peak at 547 nm, which is similar to previously reported observations²⁸ and is attributed to the NBE emission of MAPbBr₃.¹⁷

Figure 1

(a) XRD diffraction pattern of the CH₃NH₃PbBr₃ thin film deposited onto a glass substrate. (b) Corresponding photoluminescence spectrum in the 500–700 nm wavelength range. Inset shows a photograph of the deposited sample.

For the synthesis of the MAPbBr₃ nanoparticles and QDs, we prepared the MAPbBr₃ solution using equimolar amounts of MABr and PbBr₂ in DMSO. The filtered MAPbBr₃–DMSO solution was dripped onto mp-TiO₂ photoelectrodes and spin coated at a desired speed. We found that theoptimized spin-coating process led to the formation of MAPbBr₃ QDs with different sizes. Figure 2a shows a transmission electron microscopy (TEM) micrograph of TiO₂ nanoparticles coated with MAPbBr₃ QDs. The highly magnified TEM image shows that the crystalline mp-TiO₂ nanoparticles are decorated with the crystalline MAPbBr₃ QDs (Figure 2b). The average size of each QD is ~2–3 nm, as shown in Figure 2c. Figure 2c also shows a representative high-resolution TEM (HRTEM) image of the MAPbBr₃ QDs and their corresponding fast Fourier transform pattern (inset). These measurements indicate that the nanoparticles have crystalline surfaces without an amorphous layer. We also verified the composition and uniform distribution of the deposited MAPbBr₃ QDs on the mp-TiO₂ surface using scanning transmission electron microscopy (STEM). Elemental mapping was performed using energy dispersive X-ray analysis (Supplementary Information; Supplementary Figure S1). The HRTEM images show well-defined lattice spacing, and the fast Fourier transform patterns show well-defined spots that match the crystallographic parameters of the MAPbBr₃ QDs and anatase TiO₂. These results demonstrate that the nanoparticles are highly crystalline and have the same stoichiometry as the bulk.

Figure 2

Structural analysis of mp-TiO₂ coated with CH₃NH₃PbBr₃ quantum dots (~2–3 nm). (a) TEM micrograph of a CH₃NH₃PbBr₃ perovskite deposited on mp-TiO₂ nanoparticles. (b–c) Highly magnified TEM images of CH₃NH₃PbBr₃-coated TiO₂ nanoparticles. Inset of c shows afast Fourier transform pattern of CH₃NH₃PbBr₃. (d) HRTEM image of CH₃NH₃PbBr₃+mp-TiO₂.

To study the properties of the nanoparticles in detail, we synthesized different sized MAPbBr₃ perovskites by controlling the experimental conditions. Supplementary Figures S2 show typical TEM/HRTEM analyses of MAPbBr₃+mp-TiO₂ composites. The experimental techniques section discusses this in greater detail. Initially, we deposited mp-TiO₂ using the spin-coating method. This was used as the base for depositing the MAPbBr₃ nanoparticles. In a typical experiment, we varied the spin-coating speed from 3000 to 5000 r.p.m. In a typical experiment, ~10 nm MAPbBr₃ perovskite nanoparticles were produced, first by spin coating the MAPbBr₃–DMSO solution at 2500 r.p.m. (30 s) followed by an additional 30 s cycle at 3000 r.p.m. (Supplementary Figure S2). Supplementary Figure S2a–d shows the TEM/HRTEM analysis of the ~10 nm MAPbBr₃ nanoparticles. The ~6–7 nm MAPbBr₃ perovskite nanoparticles were synthesized by increasing the spin coating rate to 3000 r.p.m. (30 s) and 4500 r.p.m. (30 s) (Supplementary Figure S3). The ~4–5 nm MAPbBr₃ QDs were synthesized using the 3000 r.p.m. (10 s) and 5000 r.p.m. (45 s) spin-coating rates (Supplementary Figure S4). The effect of the QD addition was determined by recording the PL spectra for all samples. Supplementary Figure S5 shows the PL spectra of MAPbBr₃ nanoparticle/QDs of different sizes. A small red shift of the emission peak was observed as a function of size of the MAPbBr₃ perovskite QDs.^{28, 29}

Figure 3 shows the TEM/HRTEM analysis of a MAPbBr₃+mp-TiO₂ composite. For this sample, the MAPbBr₃–DMSO solution was first spin cast at 2000 r.p.m. (10 s) and then at 5000 r.p.m. (45 s). Following this, the sample was dried on a hot plate.³⁰ Figure 3a clearly shows that the mp-TiO₂ nanoparticles formed an inter-connecting nanoparticulate network. Such a network provides a low grain boundary and a fast electron transport. Figure 3b shows a representative highly magnified TEM image of a MAPbBr₃+mp-TiO₂ composite. It is clear that the TiO₂ surface is decorated with MAPbBr₃ QDs. The average size of the MAPbBr₃ QDs is ~5 nm (Figure 3c). Furthermore, the MAPbBr₃ QDs have a single phase crystalline nature, as confirmed by the HRTEM analysis (Figure 3d). We used these synthesized MAPbBr₃ nanoparticles and the QD-decorated mp-TiO₂ nanoparticles to fabricate the solar cell devices.

Figure 3

TEM analysis of a CH₃NH₃PbBr₃+mp-TiO₂ composite with ~5 nm CH₃NH₃PbBr₃ particle sizes. (a) TEM micrograph of a CH₃NH₃PbBr₃ perovskite deposited on mp-TiO₂ interconnected nanoparticles. (b–c) Highly magnified TEM images of CH₃NH₃PbBr₃-coated TiO₂ nanoparticles. (d) HRTEM image of CH₃NH₃PbBr₃ QDs, which are denoted by the yellow dotted circles.

Figure 4 shows a schematic of the typical device architectures that we used, and the energy level diagrams for the MAPbBr₃-based solar cells with spiro-MeOTAD and PTAA HTMs. In a typical PSC, the absorber layer (MAPbBr₃) is sandwiched between the ETL and the hole-transporting layer (HTL). The compact titanium oxide blocking layer (Bl-TiO₂) with or without the mesoporous scaffold (TiO₂) acts as the ETL. The HTM (spiro-MeOTAD or PTAA) and the counter electrode (Au, Ag or Al) are the key components of the HTL. The operating principle of the PSC is as follows. When the light illuminates the device the absorbing material (MAPbBr₃) absorbs the photons to generate the electron–hole pair excitons. Then, the excitons dissociate at the TiO₂/perovskite interface. The dissociated electrons flow towards the TiO₂ conduction band (CB) to FTO, whereas the holes travel through the HTM layer and reach the Au counter electrode.

Figure 4

Device configurations and energy level diagrams of CH₃NH₃PbBr₃-based perovskite solar cells used in this study. (a–b) Spiro-MeOTAD-based devices. (c–d) PTAA. The energy level values are obtained from the published literature.

The photocurrent density–voltage (J–V) characteristics of the fabricated devices were investigated under AM 1.5 G solar irradiance. The average overall photovoltaic performance is presented in Table 1. In this report, we compared two HTMs with different sizes of MAPbBr₃ perovskite as the light absorbing layer (Table 2). Supplementary Figure S6 shows J–V characteristics of the spiro-MeOTAD HTM-based PSC in the forward and reverse scan modes. The J–V measurements were carried out with a 40 ms scanning delay in the reverse (from the open-circuit voltage (V_OC) to the short-circuit current (I_SC)) and in the forward (that is, from I_SC to V_OS) modes. The spiro-MeOTAD HTM-based solar cells showed V_OC values in the range of ~0.820–0.910 V. Supplementary Figure S6 shows that all of the tested devices exhibited a V_OC of ~0.900 V. The bulk MAPbBr₃ (~10 nm particle size)-based devices show a short-circuit current density (J_SC) of 8.15 mA cm⁻², an open-circuit voltage (V_OC) of 0.888 V and a FF of 0.54, resulting in a PCE of 3.91%. The MAPbBr₃-based perovskite devices with ~7–8 nm particle sizes show a 4.22% conversion efficiency with the following parameters: J_SC=8.97 mA cm⁻², FF=0.54 and V_OC=0.873 V. By contrast, the MAPbBr₃ devices with 5–7 nm particle sizes have the following characteristics: V_OC=0.894 V, J_SC=9.79 mA cm⁻², FF=0.52, resulting in a PCE of 4.55% in the reverse scan mode. However, these devices have 3.02, 4.04 and 4.45% PCE values in forward scan direction for the 10, 7 and 5 nm, respectively (Supplementary Table S1).

Table 1 Average solar cell properties of the optimized MAPbBr₃ quantum dot-based (~5 nm) perovskite solar cells

Table 2 Average solar cell properties of the optimized MAPbBr₃ quantum dot-based (~2–3 nm) perovskite solar cells

A similar trend was observed for the PTAA-based MAPbBr₃ solar cell devices (Supplementary Table S2). All devices show V_OC values between 1.000 and 1.100 V. MAPbBr₃ solar cells without QDs show lower PCE (up to 4.40%) (Supplementary Figure S7). The 7–8 nm MAPbBr₃-based devices show the following characteristics: J_SC=9.44 mA cm⁻², FF=0.58 and η=5.73%. The 5–7 nm MAPbBr₃ nanoparticle-based samples exhibited the following characteristics: V_OC=1.082 V, J_SC=10.85 mA cm⁻², FF=0.59 and a 6.93% conversion efficiency (Supplementary Table S2).

However, all of these devices show low FF, and the high hysteresis behavior observed might be due to the thickness of the mp-TiO₂ layer.¹⁷ Therefore, we optimized the thickness of mp-TiO₂ by varying the spin-coating speed. Supplementary Figure S8 shows the cross-sectional micrographs of a mp-TiO₂+MAPbBr₃ composite deposited at various spinning rates. From the cross-sectional field emission scanning electron microscopy micrographs, we observed that the thickness of the mp-TiO₂ layer that was deposited at 2500, 3000, 4000 and 5000 r.p.m. was 533, 423, 364 and 279 nm, respectively. Furthermore, we verified the uniformity of the layer for the HTM deposition for the respective photoelectrodes. The inset micrographs show the top view of mp-TiO₂+MAPbBr₃ covered with the HTM material. These micrographs confirmed that the HTM uniformity on the mp-TiO₂+MAPbBr₃ composite increased with a higher spin-coating rate. A complete pin hole-free coverage of the HTM (spiro-MeOTAD or PTAA) was achieved at a spin coat speed of 5000 r.p.m. The HTM layer uniformity helps to increase the V_OC and FF parameters of the device.

Figure 5 shows typical J–V plots in forward and reverse scan modes for MAPbBr₃ with ~5 nm particle sizes. The FTO/Bl-TiO₂/mp-TiO₂+MAPbBr₃ (~5 nm)/spiro-MeOTAD/Au-based perovskite exhibits a PCE of 6.15% and 6.45% for the forward and reverse scan, respectively. For the standard spiro-MeOTAD-based devices, the forward scan exhibits the following characteristics: V_OC=0.941 V, J_SC=10.39 mA cm⁻², FF=0.63 and η=6.15%. The reverse scan shows similar values: PCE=6.45% (V_OC=0.928 V, J_SC=10.54 mA cm⁻² and FF=0.66). The PTAA-based devices show a 5.92 and 7.13% PCE for the forward and reverse scan modes, respectively. This sample shows high hysteresis behavior. Therefore, we optimized the conditions of the Bl-TiO₂ layer deposition and the thickness of the mp-TiO₂ layer. The Bl-TiO₂ layer was deposited by spin coating the 0.15 M solution twice followed by spin coating the 0.3 M solution once. The sample was annealed at 500 °C and treated with TiCl₄. The mp-TiO₂ layer was deposited at 5000 r.p.m. and the MAPbBr₃–DMSO solution was spin cast at 1000 (10 s) and 5000 (45 s) r.p.m., respectively. Figure 6 shows typical J–V plots of the optimized ~2–3 nm MAPbBr₃ QD devices that are based on spiro-MeOTAD and PTAA HTMs in forward and reverse scan modes. The FTO/Bl-TiO₂/mp-TiO₂+MAPbBr₃(⩽3 nm)/spiro-MeOTAD/Au-based perovskite exhibits a PCE of 7.66 and 7.61% for the forward and reverse scan modes, respectively. In case of the standard spiro-MeOTAD-based devices, the forward scan exhibits the following parameters: V_OC=0.951 V, J_SC=11.68 mA cm⁻², FF=0.69 and η=7.66%. The reverse scan shows similar characteristics: PCE=7.61% (V_OC=0.932 V, J_SC=11.50 mA cm⁻² and FF=0.71). Little improvement in the FF was observed but this did does not significantly affect the PCE. Furthermore, we studied the effect of the forward and reverse scan for the FTO/Bl-TiO₂/mp-TiO₂+MAPbBr₃/PTAA/Au-based PSCs, as shown in Figure 6. The J_SC, V_OC and FF values that were obtained from the J–V curves of the forward scan were 12.31 mA cm⁻², 1.097 V and 0.69, respectively, thus, yielding a PCE of 9.31%. By contrast, the corresponding values from the J–V curve of the forward scan were J_SC=12.05 mA cm⁻², V_OC=1.089 V and FF=0.72, thus, yielding a 9.44% conversion efficiency. The reverse scan demonstrated a drastic improvement in the FF results and in the 9.44% PCE. Overall, the forward and reverse scans exhibited almost similar efficiency, which indicates that each layer of these devices was optimized well.¹⁸ The above observations demonstrate that the V_OC was drastically improved (up to 1.097 V) due to the level of the highest occupied molecular orbital (HOMO) of PTAA (−5.2 eV) rather than spiro-MeOTAD (−5.1 eV). The PTAA-based device showed high conversion efficiency. Interestingly, the current density increased slightly (12.31 mA cm⁻²), whereas the V_OC increased drastically up to 1.097 V compared with spiro-MeOTAD. Therefore, the PTAA device shows higher conversion efficiency than the spiro-MeOTAD device. Analysis of the HOMO level of both HTMs showed that PTAA exhibits a slightly higher HOMO level. This facilitates the V_OC increase. Moreover, there is little difference among the following parameters: (1) valence band maximum of MAPbBr₃ and the HOMO level of PTAA in comparison with spiro-MeOTAD; (2) the large offset between the conduction band minimum of MAPbBr₃ and the quasi-Fermi levels of TiO_2; and (3) the possible interaction between TiO₂/MAPbBr₃/PTAA. In addition, the hole mobility of PTAA (P-type (mobility=10⁻³–10⁻² cm² V⁻¹ s))³¹ is much higher than spiro-MeOTAD (2 × 10^–4 cm² V⁻¹ s).^{32, 33, 34} Therefore, the current density also increases. Supplementary Figure S9 depicts the PCE dependence on MAPbBr₃ particle size with different HTMs. Figure 7 shows the air stability of the fabricated perovskite devices. For the stability evaluation, we stored the devices in an Al foil-covered Petri dish in air at room temperature without encapsulation. The V_OC and J_SC values of both devices were stable for up to 2500 h, after which a slight decline was observed.

Figure 5

J–V curves of the forward and reverse bias sweeps and the respective J–V curves for spiro-MeOTAD and PTAA that have a CH₃NH₃PbBr₃ perovskite absorber layer of ~5-nm-sized particle. J–V curves were measured using forward and reverse scans with 10 mV voltage steps and 50 ms delay times under AM 1.5 G illumination.

Figure 6

J–V curves of forward and reverse bias sweeps and respective J–V curves for spiro-MeOTAD and PTAA that have a CH₃NH₃PbBr₃ perovskite (~2–3 nm) absorber layer. J–V curves were measured using forward and reverse scans with 10 mV voltage steps and 50 ms delay times under AM 1.5 G illumination.

Figure 7

Stability measurements of CH₃NH₃PbBr₃ nanoparticle-based perovskite solar cells. (a) FTO/Bl-TiO₂/mp-TiO₂+CH₃NH₃PbBr₃/spiro-MeOTAD/Au. (b) FTO/Bl-TiO₂/mp-TiO₂+CH₃NH₃PbBr₃/PTAA/Au.

Figure 8a shows a typical cross-sectional field emission scanning electron microscopy image of the fabricated device. The appropriate thickness of each layer confirms that the optimized device has 250 nm of mp-TiO₂ with ⩽3 nm of MAPbBr₃ QDs and 180 nm of the HTM layer and 80 nm of the gold contact. The highly magnified TEM image shows that the surface of mp-TiO₂ is well coated with the MAPbBr₃ particles (Figure 2c). Figure 8b shows the J–V curve of the champion cell that was recorded for the MAPbBr₃ perovskite. This layer was deposited with little variation according to our previous method. The MAPbBr₃–DMSO solution was dripped onto mp-TiO₂, allowed to soak for ~30 s, and then spin coated. The spin-coating speed was maintained at 1000 r.p.m. and at 5000 r.p.m. for 10 and 40 s, respectively. The deposited sample was dried on a hot plate to evaporate the solvent and to crystallize MAPbBr₃. The dried mp-TiO₂+MAPbBr₃ photoelectrodes were used for the PTAA deposition. The PTAA HTM layer was deposited at 3500 r.p.m. for 30 s and 80 nm gold fingers were deposited using thermal evaporation. For the champion cell, the highest PCE of 11.41% was achieved because of the high V_OC=1.110 V and the high current density of 14.07 mA cm⁻². It was observed that the FF=0.73 was also much higher than for the other samples tested. This enhancement was primarily due to the lower back reaction, lower recombination rate, high FF and high current density. To confirm this observation, we also recorded the incident photon-to-electron conversion efficiency (IPCE) data. Figure 8c shows the IPCE spectra as a function of the wavelength of the champion cell. The IPCE spectra provide detailed information on the effective light harvesting capabilities of the perovskite-based solar cells. The photocurrent generation starts at ~550 nm, which is in agreement with the band gap of CH₃NH₃PbBr₃, and reaches up to ~82% IPCE in the visible spectrum.

Figure 8

Solar cell properties of the champion cell with a FTO/Bl-TiO₂/mp-TiO₂+CH₃NH₃PbBr₃/PTAA/Au device configuration. (a) Cross-sectional field emission scanning electron microscopy image of a fabricated perovskite solar cell device. (b) Photocurrent density–voltage (J–V) curve and (c) incident power conversion efficiency (IPCE) spectrum of the champion cell.

Conclusions

In summary, we synthesized methylammonium lead bromide MAPbBr₃ QDs via a DMSO solvent-based method and used these QDs to fabricate PSCs with different HTMs. Fabrication of MAPbBr₃ perovskite of different particle sizes was controlled by the experimental conditions. The photovoltaic performance was significantly dependent on the size of the MAPbBr₃ nanoparticles and QDs. The best conversion efficiency of 11.46% (V_OC=1.110 V, J_SC=14.07 mA cm⁻²) was achieved with PTAA as the HTM and a 0.73 FF for the ⩽3 nm MAPbBr₃ QDs. This solar cell performance enhancement was attributed to the MAPbBr₃ QDs having sizes ⩽3 nm. This result enables us to elucidate the perovskite operating mechanism. The hysteresis study revealed that the MAPbBr₃-based cells show a very low tolerance towards the ~250 nm mp-TiO₂ thickness. We believe that our method can be applied to achieve stable planar hybrid PSCs. The implementation of MAPbI₃ and MAPbBr₃ QDs will also open new opportunities for the development of efficient PSCs. Such studies, based on different HTMs, are currently underway in our laboratory.